Abstract
Test-analysis models are used in the validation of the nite element models of spacecraft structures. Here, a probabilistic approach is used to assess the robustness of a system equivalent reduction expansion process based testanalysis model when experimental and analytical modes contain different levels of inaccuracy. The approach is applied to three spacecraft models, and Monte Carlo simulations were used to determine the sensitivity of the normalized cross-orthogonality check to the system equivalent reduction expansion process reduced matrix. The effect of parameters used in this reduction and the amount of inaccuracies that can be tolerated in the modes before failing the normalized cross-orthogonality check were also determined. The results show that the probability to pass the normalized cross-orthogonality check is highly determined by the number of modes used in the reduction. The relation between capability of the nite element models to predict the frequency-response function and the quality of the model validation determined using normalized cross-orthogonality check is also investigated, and it is observed that the quantities are not always correlated. This study also shows that the sensor locations can be optimally chosen using the system equivalent reduction expansion process reduced mass matrix, and this can increase the probability to pass the normalized cross-orthogonality check.